On Fr, 2019-03-29 at 16:50 +0300, Dan Carpenter wrote: > External E-Mail > > > Thanks, I feel like I understand better now. > > Sorry, I don't want to be a jerk and I'm not going to complain again > about this patchset if you resend it with the other stuff I mention > fixed. > > But to me it feels like the API could be improved slightly if you > first created the adapter with a name and then linked to > it. Creating > the adapter as a side effect of linking the first audio component > feels > like it seems helpful but it's going to be awkward. Hmm, I see. But I'm not sure if I can call snd_pcm_new() on an already registered sound adapter. > > I'm also concerned about the locking around the adpt_list. Is this > something ordinary users can trigger? Is it crash if I set my fuzz > tester to start randomly connecting and disconnecting like crazy? > Well, strictly speaking yes. Because it is user space. But normally, the driver is meant for engineered systems like a car (see Automotive Grade Linux). Here you will have a systemd unit that sets things up at system startup time. I don't expect having to deal with multiple user space applications that race about the sound adapters. But I will keep that in mind. thanks, Chris > regards, > dan carpenter > > _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
